Re: Extension/Extensibility examples in W3C Specifications

Le 04 mai 2004, à 11:20, Dominique Hazaël-Massieux a écrit :
>>  One possibility Lynne and I discussed is
>> to just use "extensibility" and not use, or define, "extensions"
>
> I think that extensibility is indeed a better term to describe what we
> want to talk about in SpecGl ; that said, given that the term
> "extensions" seems to give us troubles, it would be good to have the 
> set
> of definitions that the term "extension" covers, so as to reduce the
> ambiguity later on.

We have really to focus on the definitions and their meanings.

Everything in the world is extensible, or almost. It's particulary true 
for a technology.
	A technology is de facto extensible and can have extensions.
	Therefore, the extensibility nature of a technology is true, all the 
time.

Now we are working in the context of the W3C, a *standard 
organization*. I really would like to stress this point. A technology 
at W3C is something which has been, in the best case,  built with the 
participation of W3C Members and public participation and a consensus.

	Technology LoveML is defined. You can be conformant or not to this 
technology LoveML.

This mean that you respect certain criteria when you implement the 
technology. These criteria are defined in a conformance section. The 
LoveML WG has foresee that some people would like to develop specific 
implementations with more features, more possibilities, but very 
particular to a product or a specific market.

They decide to make LoveML extensible and so to create an extensibility 
mechanism. They would like to avoid the mess between different 
products, they would like to avoid to have interoperability problems. 
So they develop a mechanism which makes possible to create extensions 
to LoveML in a harmonious way. They define rules for that.

	Now let's dig a bit with a real example:

* XHTML 1.1 is an implementation of XHTML Modularization.
* XHTML Modularization has a very strict framework to create other 
XHTML Mod Modules in a conformant way.

It means you can designed for example a PoetryML Module for XHTML 1.1 
without screwing XHTML 1.1, Let's call the final thing: "XHTML Poetry" 
which is XHTML 1.1 + PoetryML Module.

The results are:

1. You can NOT be conformant to XHTML Poetry per W3C Rules.
2. An external organization may have defined Conformance Rules to XHTML 
Poetry (fine)
3. Your PoetryML Module is conformant to regards to the conformance 
Rules of XHTML Modularization.

Conclusion:
	Extensibility Framework gives the possibility to avoid 
interoperability problems and to design extensions to spec which are 
conformant to this extensibility framework.
	It doesn't mean that the technology + its extension define a new 
conformance set.


===> It's my way of seeing things and it's the whole purpose of a 
standard organization, IMHO. To have a minimal set of requirements 
which are interoperable.

Open to debate ;)

-- 
Karl Dubost - http://www.w3.org/People/karl/
W3C Conformance Manager
*** Be Strict To Be Cool ***

Received on Tuesday, 4 May 2004 17:15:53 UTC